Subject : IOTA will provide the next generation of models and algorithms to advance our understanding of IoT analytics.
Project Coordinators : Albert Bifet, Silviu Maniu
Project Members :Nikolaos Tziortziotis (PostDoc) starting Nov 2018, Maroua Bahri (PhD Student)
Host Laboratory : LTCI, LRI
DigiCosme Funding : 2016/2019
The Internet of Things (IoT), the large network of physical devices that extends beyond the typical computer networks, will be creating a huge quantity of Big Data streams in real time in the near future. The realization of IoT depends on being able to gain the insights hidden in the vast and growing seas of data available. Since current approaches do not scale to Internet of Things (IoT) volumes, new systems with novel mining techniques are necessary due to the velocity, but also variety, and variability, of such data.
This IoT challenging setting needs algorithms that use an extremely small amount (iota) of time and memory resources, and that are able to adapt to changes while not stopping the learning process. These algorithms should be distributed to allow them to run on top of Big Data infrastructures. How to do this accurately in a fully automatic, and transparent elastic, real-time, system is going to be the main challenge for IoT analytics systems in the near future.
In this IoT scenario, high-performing ensemble setups such as online bagging, leveraging bagging and random forests are the state-of-the-art. Deep neural networks are becoming increasingly popular, owing in part to the proliferation of interest and oft-advertised successes in deep learning. These algorithms can learn incrementally, but they have so far proved too sensitive to hyper-parameter options and initial conditions to be considered for the IoT data stream setting.
IOTA will push into the next generation of models and algorithms to advance our understanding of IoT analytics. To achieve this goal, IOTA aims to employ a combination of : 1/ Adaptive Never-Ending Learning Systems, 2/ Distributed Big Data Systems.
The new proposed techniques are specifically designed for industrial-sized programs; the high quality software produced in IOTA is going to be open source, so that industry practitioners and academic researchers can benefit from these new advances.
Scientific productions :
- Maroua Bahri, Albert Bifet, Silviu Maniu, Rodrigo Fernandes de Mello, Nikolaos Tziortziotis. Compressed k-Nearest Neighbors Ensembles for Evolving Data Streams.In the 24th European Conference on Artificial Intelligence (ECAI), 2020, Santiago de Compostela, Spain.
- Maroua Bahri, Bernhard Pfahringer, Albert Bifet, Silviu Maniu. Efficient Batch-Incremental Classification for Evolving Data Streams. In the Symposium on Intelligent Data Analysis (IDA), 2020, Lake Constance, Germany.
- Jesus L. Lobo, Javier Del Ser, Albert Bifet, Nikola Kasabov: Spiking Neural Networks and online learning: An overview and perspectives. Neural Networks 121: 88-100 (2020)
- Jesus L. Lobo, Izaskun Oregi, Albert Bifet, Javier Del Ser: Exploiting the stimuli encoding scheme of evolving Spiking Neural Networks for stream learning. Neural Networks 123: 118-133 (2020)
- Bogdan Cautis, Silviu Maniu, Nikolaos Tziortziotis: Adaptive Influence Maximization. KDD 2019: 3185-3186
- Heitor Murilo Gomes, Jesse Read, Albert Bifet, Jean Paul Barddal, João Gama: Machine learning for streaming data: state of the art, challenges, and opportunities. SIGKDD Explorations 21(2): 6-22 (2019)
- Jean Paul Barddal, Fabrício Enembreck, Heitor Murilo Gomes, Albert Bifet, Bernhard Pfahringer: Merit-guided dynamic feature selection filter for data streams. Expert Syst. Appl. 116: 227-242 (2019)
- Rodrigo Fernandes de Mello, Yule Vaz, Carlos Henrique Grossi Ferreira, Albert Bifet: On learning guarantees to unsupervised concept drift detection on data streams. Expert Syst. Appl. 117: 90-102 (2019)
- Rodrigo Fernandes de Mello, Chaitanya Manapragada, Albert Bifet: Measuring the Shattering coefficient of Decision Tree models. Expert Syst. Appl. 137: 443-452 (2019)
- Robert Anderson, Yun Sing Koh, Gillian Dobbie, Albert Bifet: Recurring concept meta-learning for evolving data streams. Expert Syst. Appl. 138 (2019)
- Abhik Ray, Lawrence B. Holder, Albert Bifet: Efficient frequent subgraph mining on large streaming graphs. Intell. Data Anal. 23(1): 103-132 (2019)
- Minh-Huong Le Nguyen, Heitor Murilo Gomes, Albert Bifet: Semi-supervised Learning over Streaming Data using MOA. BigData 2019: 553-562
- Heitor Murilo Gomes, Rodrigo Fernandes de Mello, Bernhard Pfahringer, Albert Bifet: Feature Scoring using Tree-Based Ensembles for Evolving Data Streams. BigData 2019: 761-769
- Dihia Boulegane, Albert Bifet, Giyyarpuram Madhusudan: Arbitrated Dynamic Ensemble with Abstaining for Time-Series Forecasting on Data Streams. BigData 2019: 1040-1045
- Dihia Boulegane, Nedeljko Radulovic, Albert Bifet, Ghislain Fiévet, Jimin Sohn, Yeonwoo Nam, Seojeong Yu, Dong-Wan Choi: Real-Time Machine Learning Competition on Data Streams at the IEEE Big Data 2019. BigData 2019: 3493-3497
- Heitor Murilo Gomes, Jesse Read, Albert Bifet: Streaming Random Patches for Evolving Data Stream Classification. ICDM 2019: 240-249
- Heitor Murilo Gomes, Albert Bifet, Philippe Fournier-Viger, Jones Granatyr, Jesse Read: Network of Experts: Learning from Evolving Data Streams Through Network-Based Ensembles. ICONIP (1) 2019: 704-716
- Luis Eduardo Boiko Ferreira, Heitor Murilo Gomes, Albert Bifet, Luiz S. Oliveira: Adaptive Random Forests with Resampling for Imbalanced data Streams. IJCNN 2019: 1-6
- Guilherme Weigert Cassales, Hermes Senger, Elaine Ribeiro de Faria, Albert Bifet: IDSA-IoT: An Intrusion Detection System Architecture for IoT Networks. ISCC 2019: 1-7
- Albert Bifet, Ricard Gavalda, Geoff Holmes, Bernhard Pfahringer. Machine Learning from Data Streams. MIT Press, 2018.
- Albert Bifet, Jesse Read: Ubiquitous Artificial Intelligence and Dynamic Data Streams. DEBS 2018: 1-6
- Andrian Putina, Steven Barth, Albert Bifet, Drew Pletcher, Cristina Precup, Patrice Nivaggioli, Dario Rossi: Unsupervised real-time detection of BGP anomalies leveraging high-rate and fine-grained telemetry data. INFOCOM Workshops 2018: 1-2
- Andrian Putina, Dario Rossi, Albert Bifet, Steven Barth, Drew Pletcher, Cristina Precup, Patrice Nivaggioli: Telemetry-based stream-learning of BGP anomalies. Big-DAMA@SIGCOMM 2018: 15-20
- Nicolas Kourtellis, Gianmarco De Francisci Morales, Albert Bifet: Large-Scale Learning from Data Streams with Apache SAMOA. CoRR abs/1805.11477 (2018)
- Jacob Montiel, Jesse Read, Albert Bifet, Talel Abdessalem: Scikit-Multiflow: A Multi-output Streaming Framework. CoRR abs/1807.04662 (2018)
- Heitor Murilo Gomes, Jean Paul Barddal, Fabrício Enembreck, Albert Bifet: A Survey on Ensemble Learning for Data Stream Classification. ACM Comput. Surv. 50(2): 23:1-23:36 (2017)
- Diego Marron, Jesse Read, Albert Bifet, Nacho Navarro: Data stream classification using random feature functions and novel method combinations. Journal of Systems and Software 127: 195-204 (2017)
- Heitor Murilo Gomes, Albert Bifet, Jesse Read, Jean Paul Barddal, Fabrício Enembreck, Bernhard Pfharinger, Geoff Holmes, Talel Abdessalem: Adaptive random forests for evolving data stream classification. Machine Learning 106(9-10): 1469-1495 (2017)
- Diego Marron, Eduard Ayguadé, José R. Herrero, Jesse Read, Albert Bifet: Low-latency multi-threaded ensemble learning for dynamic big data streams. BigData 2017: 223-232
- Jacob Montiel, Albert Bifet, Talel Abdessalem: Predicting over-indebtedness on batch and streaming data. BigData 2017: 1504-1513
- Albert Bifet: Classifier Concept Drift Detection and the Illusion of Progress. ICAISC (2) 2017: 715-725
- Pierre-Xavier Loeffel, Albert Bifet, Christophe Marsala, Marcin Detyniecki: Droplet Ensemble Learning on Drifting Data Streams. IDA 2017: 210-222
- Albert Bifet, Jiajin Zhang, Wei Fan, Cheng He, Jianfeng Zhang, Jianfeng Qian, Geoff Holmes, Bernhard Pfahringer: Extremely Fast Decision Tree Mining for Evolving Data Streams. KDD 2017: 1733-1742